Site-specific ¹⁹F NMR chemical shift and side chain relaxation analysis of a membrane protein labeled with an unnatural amino acid

Abstract

Site-specific ¹⁹F chemical shift and side chain relaxation analysis can be applied on large size proteins. Here, one-dimensional ¹⁹F spectra and T₁, T₂ relaxation data were acquired on a SH3 domain in aqueous buffer containing 60% glycerol, and a nine-transmembrane helices membrane protein diacyl-glycerol kinase (DAGK) in dodecyl phosphochoine (DPC) micelles. The high quality of the data indicates that this method can be applied to site-specifically analyze side chain internal mobility of membrane proteins or large size proteins.

Introduction

Nuclear magnetic resonance (NMR) spectroscopy can provide a detailed, site-specific description of internal motions in proteins.¹ Most of protein side chain internal motions are analyzed through ¹³C or ²H relaxation. However, acquisition of relaxation data is hampered as the sizes of the proteins increases, mainly due to difficulties of resonance assignment and relative low sensitivity of side chain isotopes.

¹⁹F NMR spectroscopy has been widely applied to investigate protein structure change and dynamics, and ¹⁹F NMR motional analysis of several large size proteins were reported, including a 100 kDa citrate synthase² and a 210 kDa trimeric pyruvate kinase,³ despite low signal-to-noise ratio and lack of ¹⁹F peak assignment.

Recently, an unnatural amino acid, tri-fluoromethyl-phenylalanine, was introduced to specific sites of a protein for further ¹⁹F NMR studies.^4–6 Briefly, the tfmF was recognized by an artificially evolved tRNA synthetase (RS) and linked to a tRNA containing an amber nonsense anticodon (CUA), catalyzed by the same RS. Then, tfmF can be specifically incorporated into a protein at the amber nonsense codon (TAG) in the protein's coding DNA sequence. Using this site-specific ¹⁹F incorporation method, ¹⁹F chemical shift analysis was shown to be possible for a large protein.⁴

Here, site-specifically incorporated tfmF and ¹⁹F NMR were applied to study side chain internal mobility of large size proteins. High quality ¹⁹F chemical shift, T₁, T₂ relaxation data from samples of human vinexin SH3 domain in aqueous buffer containing 60% glycerol and a nine-transmembrane helix membrane protein diacyl-glycerol kinase (DAGK) in DPC micelles indicate that side chain internal mobility analysis can be achieved for large molecular systems.

Results and Discussion

The tfmF was specifically introduced at Phe7 site of human vinexin SH3 domain for further ¹⁹F NMR studies. As the tfmF was site-specifically introduced into Phe7 of SH3, only one peak was observed and assigned to Phe7 in one-dimensional ¹⁹F spectrum [Fig. 1(A,D)]. The straightforward resonance assignment makes it convenient for chemical shift and relaxation analysis. With the absence or presence of a proline-rich peptide ligand P868 (GEVPPPRPPPPEE), which was known to bind to SH3 domain in high affinity,⁷ different ¹⁹F chemical shift [Fig. (1A,D)], T₁ [Fig. 1(B,E)], and T₂ relaxation values [Fig. 1(C,F)] were observed. Observed chemical shift changes are consistent with previous reports that Phe7 is located in binding pocket for P868 in human vinexin SH3 domain.⁷ About a 1.5-fold increase in T₁ (1496.00 ± 286.71 vs. 1095.70 ± 68.72 ms) and a 10-fold increase in T₂ (1088.04 ± 217.47 vs. 89.24 ± 5.84 ms) were observed upon ligand binding. According to relaxation theory, ratio of T₁/T₂ is a function of internal mobility,^8–10 and a T₁/T₂ ratio decrease or remarkable T₂ increase indicate a liberated internal motion.¹¹ It has been known that relaxations of ¹⁹F spins in tfmF are dominated by chemical shift anisotropy,¹⁰ due to lack of directly bonded protons and consequent dipolar couplings. Taking advantage of fast rotational motion of CF3- group about the C—C bond to the aromatic ring, ¹⁹F T₁ value of free tfmF (T₁ = 1.47 ± 0.06 s) is about 200-fold shorter than 4-F-Phe (T₁ = 352 ± 12 s), in which ¹⁹F spin is directly bonded to aromatic ring.¹² Therefore, application of tfmF for ¹⁹F NMR chemical shift and relaxation studies is much superior to 4-F-Phe. It is also noticed that measured T₁, T₂ relaxation values of Phe7-tfmF-SH3 at presence of ligand P868 are in similar level as T₁, T₂ value of free tfmF (T₁ = 1470.63 ± 62.10 ms, T₂= 969.55 ± 199.38 ms, data not shown). Explanations for the observations can be that previously restrained Phe7 side chain reorientation motion of human vinexin SH3 domain was liberated, when ligand P868 binds to the binding pocket.

Figure 1.

¹⁹F chemical shift and T₁, T₂ relaxation analysis of the Phe7-tfmF-SH3 domain in aqueous buffer. Different ¹⁹F chemical shifts (A, D) and T₁ relaxation (B, E), T₂ relaxation values (C, F) are observed in the absence (A, B, C) or presence (D, E, F) of a peptide ligand P868.

To assess whether ¹⁹F NMR side chain mobility studies can be applied for large proteins, samples of SH3 in 60% glycerol were prepared to mimic a protein with short correlation time, for example, a large size protein in aqueous buffer. A pulse field gradient spin echo NMR pulse sequence was applied to measure water diffusion constant in this sample.^13,14 In Figure 2A, diffusion coefficient of water in sample of Phe7-tfmF-SH3 in aqueous buffer containing 60% glycerol was measured (D_s = 2.289 × 10⁻¹⁰ m²/s, Fig. 2A). According to Eq. (4) in experimental methods, a small protein (M_w′) in high viscosity condition (D_s′ η′) can mimic a large size protein in aqueous buffer (M_w, D_s, η). We have known molecular weight of human vinexin SH3 domain (7.4 kDa) and diffusion coefficient of water at 298 K (2.299 × 10⁻⁹ m²/s).¹⁵ Therefore, a molecular weight of 7500 kDa [(22.99/2.289)³ × 7.4 kDa] was calculated to have the same correlation time as human vinexin SH3 domain (M_w = 7.4 kDa) in aqueous buffer containing 60% glycerol.

Figure 2.

Diffusion coefficient measurements of water (A), ¹⁹F chemical shift (B) and T₁ relaxation (C) and T₂ relaxation analysis (D) from the sample of Phe7-tfmF-SH3 in aqueous buffer containing 60% glycerol.

¹⁹F NMR spectrum (Fig. 2B) and T₁ (Fig. 2C), T₂ relaxation data (Fig. 2D) were also collected for human vinexin SH3 domain in aqueous buffer containing 60% glycerol. Despite broad line-width and small T₂ relaxation values, decent spectra and quantitative T₁, T₂ relaxation values were obtained even in a high viscosity. These data strongly indicate that ¹⁹F chemical shift and side chain internal mobility can be analyzed for large size globular proteins with molecular weights up to 7500 kDa, in aqueous buffer.

TfmF was also incorporated at Phe31 site of a membrane protein DAGK (Phe31-tfmF-DAGK) in DPC micelles for further ¹⁹F NMR studies. Trimeric DAGK contains total nine transmembrane helices and molecular weight of DAGK/DPC micelles was estimated to be around 100 kDa at 45 °C.¹⁶¹⁹F chemical shift, and T₁, T₂ relaxation analysis of DAGK/DPC were shown in Figure 3. Decent ¹⁹F spectrum (Fig. 3A) and quantitative T₁ (Fig. 3B), T₂ relaxation (Fig. 3C) verified that this ¹⁹F NMR data can be acquired from a large membrane protein.

Figure 3.

One-dimensional ¹⁹F chemical shift (A), T₁ relaxation (B) and T₂ relaxation analysis (C) of a membrane protein DAGK labelled at position 31 with tfmF in DPC micelles (Phe31-tfmF-DAGK/DPC).

In a summary, ¹⁹F chemical shift and relaxation analysis of SH3 in 60% glycerol indicate that this method can be applied on large size proteins. ¹⁹F NMR data of DAGK/DPC verify that this site-specific side chain chemical shift and internal mobility analysis provide a powerful tool to study molecular mechanisms of large size proteins, especially membrane proteins in detergent micelles.

Materials and Methods

Protein production and ¹⁹F site-specific labeling

Plasmid pBAD-His₆-SH3-TAG (a pBAD vector containing SH3 domain with its 7th codon mutated to amber stop codon TAG) and pDule-tfmF (containing DNA sequences coding tRNA_CUA and tfmF-specific aminoacyl-tRNA synthetase) were co-transformed into Escherichia coli (E. coli) host cell TOP10 with the presence of antibiotics. Protein over-expression was achieved with presence of 1 mM¹⁹F-tfmF in rich medium, similar as previous reports.^4,5 Over-expressed Phe7-tfmF-SH3 was purified using Ni²⁺-NTA affinity and size-exclusive chromatography. Final yield of about 1 mg Phe7-tfmF-SH3 was obtained from every 500 mL culture.

The 31st codon (Phe) of DAGK protein was mutated to amber stop codon and Phe31-tfmF-DAGK was over-expressed in E. coli in similar way as described above. Over-expressed Phe31-tfmF-DAGK was purified in DPC micelles as described previously.¹⁶

¹⁹F NMR

All ¹⁹F NMR spectra were acquired at 298 K, in a Bruker Avance 400 MHz spectrometer. Data were processed with an exponential window function (10 Hz line-broadening) using TopSpin 3.1. ¹⁹F chemical shifts were referenced to an internal compound of trifluoro-acetic acid (TFA, −75.39 ppm).

Protein samples for ¹⁹F NMR analysis were in final concentration of 0.2 mM, including Phe7-tfmF-SH3 or Phe31-tfmF-DAGK/DPC in 500 μL aqueous buffer of 50 mM Na₂HPO₄/NaH₂PO₄, pH 6.5. Another Phe7-tfmF-SH3 sample was prepared in the same aqueous buffer containing 60% glycerol.

Side chain ¹⁹F T₁ relaxation data were collected with eight delay times (50, 100, 200, 500, 800, 1000, 1500, 2000 ms) using a standard Bruker 1D inverse-recovery pulse sequence. ¹⁹F T₂ relaxation data were collected with 8 delay times (100, 150, 200, 400, 600, 800, 1200, and 1600 ms) using a standard Bruker 1D Carr-Purcel-Meiboom-Gill pulse sequence. Resonance intensities in relaxation experiments were measured and fit to an exponential function.

All ¹⁹F NMR spectra were acquired with 5 s acquisition delays. 64 scans were accumulated for one-dimensional ¹⁹F chemical shift analysis and 1024 scans were accumulated for T₁ and T₂ data acquisition. Therefore, totally 6 min are required for 1D NMR acquisition, while 1.5 h are required to collect each relaxation data point.

Diffusion coefficient measurement and molecular weight calculation

A pulsed field gradient-based longitudinal eddy delay pulse sequence was applied to measure water diffusion coefficient in samples of Phe7-tfmF-SH3 in aqueous buffer containing 60% glycerol. The experiments were conducted at 298 K, in a 500 MHz Bruker Avance spectrometer. A series of 1D data were collected with different delay duration of 13.184, 52.737, 118.66, 210.95, 329.61, 474.64, 646.03, and 843.80 ms. Data were processed using exponential window function and intensities of H₂O signal (4.680 ppm) were measured using Bruker xwinnmr software. Diffusion constant was regressed from intensity decay with different delay time.

The Stokes-Einstein Equation was applied to derive the hydrated radius or the molecular weight of a globular protein:

(1)

η – viscosity of the solution R_s – sphere radius of the molecule

(2)

Assuming a constant partial specific volume V, protein's molecular weight (M_w) is proportional to cubic of hydrated radius R_s³. Replacing hydrated radius R_s with molecular weight M_w, we can reach another equation:

(3)

Therefore, assuming the same correlation time at 298 K, molecular weight of a large size protein in aqueous buffer (M_w, D_sη) can be a mimic of a small protein (M_w′) in high viscosity condition (D_s′ η′):

(4)

D_s′ and D_s are the water diffusion coefficients in aqueous sample and an aqueous buffer in high viscosity, respectively.

Glossary

Abbreviations:

DAGK

diacylglycerol kinase

DPC

dodecyl phosphocholine

NMR

nuclear magnetic resonance

tfmF

p-trifluoromethyl-phenylalanine.